Output details
11 - Computer Science and Informatics
University of Salford
An empirical comparison of cost-sensitive decision tree induction algorithms
<24>There are many publications on cost-sensitive learning, each espousing a different algorithm with authors claiming to outperform other selected algorithms, and using different methodologies. This paper is significant in that it is an independent comparison using the same methodology and data. The insights presented are valuable for those selecting and applying cost-sensitive methods. For example, the insights have been used in a KTP that applied cost -sensitive induction for risk assessment in sub-prime lending, which is helping to improve social and financial inclusion and which was selected by ESRC as an exemplar of the impact from their funding (see http://www.esrc.ac.uk/publications/videos/creating-impact.aspx).